One institution that hasn't been mentioned is conference proceedings. I don't know why but mathematics generally seems to take a more casual approach to these than engineering. Whereas mathematics conferences tend to expect the talks to ripen into carefully refereed publications afterwards, whether together in a proceedings or separately in various journals, engineering conferences tend to insist on issuing a bound proceedings with 4-10 pages per paper, leaving it up to the authors to decide whether they want to submit a more polished version to a journal later. For both mathematics and engineering a conference may invite a subset of the papers for a special issue of a suitable journal. This difference is sensitive to the needs and circumstances of publication. Ostensibly the primary purpose of publication is dissemination, with author kudos supposedly secondary. Lately the latter has been badly skewing the former, with conferences seemingly worrying as much about appointments and promotions as about dissemination. This may well be a side effect of the web, whose search engines support associative retrieval of polished articles and daily blogs alike, solving the access problem without addressing the evaluation problem. This technological revolution is transforming the publication world faster than universities, libraries, and publishers can follow in real time. Appointments and promotions have until recently been mired in the tradition of relying on refereed journal publications in strong preference to conference publications. Libraries continue to follow the taste of deans in preferring to archive journals over conference proceedings, with the result that at least pre-web articles in conference proceedings are inaccessible to the clients of many libraries. And publishers seem to have a certain inertia that makes them slow to adapt their processes to the outgoing tide of publication costs, an inertia that strands them on the rocks of their expensive old methods. This is all changing, slowly but inevitably. Engineering deans are becoming more willing to equate at least flagship conference publications with journals. Search engines are making libraries less relevant for current material, while the ongoing digitization of older material is starting to make basement stacks less relevant. And authors, editors, referees, and libraries are forcing the collective hand of the publishers by avoiding the most expensive. In this disruptive scenario the potential exists for conferences to assume more of the role of journals. The effect of journal refereeing by itself is achieved for conferences with two mechanisms: refereeing (supposedly quicker and less careful than for journals), and limited capacity at the top---flagship conferences have acceptance rates of 20-40%, forcing the overflow into lesser conferences. Whereas in the past appointments and promotions were judged on the fact of journal acceptance in combination with the assessment of their quality by peers and seniors, these two criteria are now joined by a third: the quality of the conferences that accepted the candidate's papers. Historically journal quality while a factor took a back seat to mere acceptance; today it is made more important by the need to justify the supposedly less careful refereeing and certainly hastier preparation of conference papers. If this trend towards attaching more importance to conference publication is where we're all headed, it will happen in engineering before it happens in mathematics for the simple reason that engineering promotes conference publication more strenuously than does mathematics. Vaughan